Skip to content
This repository was archived by the owner on Mar 20, 2023. It is now read-only.

Bug fix for the GPU execution of BBP model: issue with mechanism data offset #590

Merged
pramodk merged 3 commits into from
Jul 19, 2021
Merged

Bug fix for the GPU execution of BBP model: issue with mechanism data offset #590

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4e04709
Bug fix for the GPU execution of BBP model: issue with mechanism data…
pramodk Jul 16, 2021
2e168d5
Rely on acc_deviceptr (reliable) than manually calculating offsets (f…
pramodk Jul 16, 2021
55b4708
fix clang-format
pramodk Jul 16, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 2 additions & 6 deletions coreneuron/gpu/nrn_acc_manager.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,7 +153,6 @@ void setup_nrnthreads_on_device(NrnThread* threads, int nthreads) {
NrnThreadMembList* d_last_tml;

bool first_tml = true;
size_t offset = 6 * ne;

for (auto tml = nt->tml; tml; tml = tml->next) {
/*copy tml to device*/
Expand Down Expand Up @@ -186,13 +185,10 @@ void setup_nrnthreads_on_device(NrnThread* threads, int nthreads) {
int is_art = corenrn.get_is_artificial()[type];
int layout = corenrn.get_mech_data_layout()[type];

offset = nrn_soa_padded_size(offset, layout);

dptr = d__data + offset;

// get device pointer for corresponding mechanism data
dptr = (double*) acc_deviceptr(tml->ml->data);
acc_memcpy_to_device(&(d_ml->data), &(dptr), sizeof(double*));

offset += nrn_soa_padded_size(n, layout) * szp;

if (!is_art) {
int* d_nodeindices = (int*) acc_copyin(tml->ml->nodeindices, sizeof(int) * n);
Expand Down